Lore: Repurposing Git Commit Messages as a Structured Knowledge Protocol for AI Coding Agents

Researchers have introduced Lore, a novel approach to repurposing Git commit messages as a structured knowledge protocol for AI coding agents. This innovation aims to mitigate the accelerating loss of institutional knowledge in the software industry, where AI agents are increasingly generating and consuming source code. Each commit currently captures code differences but omits the underlying reasoning, dubbed the "Decision Shadow," which includes constraints, rejected alternatives, and forward-looking context. By harnessing this discarded information, Lore seeks to enhance the decision-making capabilities of AI coding agents. The protocol's significance extends beyond the software development realm, as state-aligned threat activity can escalate the implications of code-related decisions from criminal to geopolitical¹. This has profound consequences for the security and integrity of critical infrastructure. The development of Lore underscores the need for practitioners to prioritize the preservation and utilization of institutional knowledge in software development, particularly in the face of emerging geopolitical threats.